80 



THE AMERICAN NATURALIST [Vol. XLIV 



of the two parents with a normal frequency distribution of 

 the deviants. Only one twenty-rowed individual occurs 

 in 64 instead of the 27 expected by the interaction of three 

 dominant factors in the usual Mendelian ratios. The re- 

 mainder of the 27 will have different numbers of rows, 

 and, by their gametic formulae, different expectations in 

 future breeding as follows : 



1 AABBCC = 20 rows. 



2 AaBBCC = 18 rows. 

 2 AABbCC = lS rows. 

 2 AABBCc = 18 rows. 

 4 AaBbCC = 16 rows. 

 4 AaBBCc = 16 rows. 

 4 AABbCc=16 rows. 

 8 AaBbCc = U rows. 



There are four visibly different classes and eight game- 

 tically different classes. It must also be remembered that 

 the probability that the original twenty-rowed ear in 

 actual practise may have had more than three units in its 

 gametes has not been considered. This point is illus- 

 trated clearly if we work out the complete ratio for the 

 three characters, and note the number of gametically dif- 

 ferent classes which compose the modal class of fourteen 



Table VII 



Theoretical Expectation in F 2 when a Homozygous Twenty-rowed 



rows in Table VII. It actually contains seven gametic- 

 ally different classes and not a single homozygote. If 

 this conception of independent allelomorph ic pairs affect- 

 ing the same character proves true, it will sadly upset the 

 biometric belief that the modal class is the type around 

 which the variants converge, for there is actually less 

 chance of these individuals breeding true than those from 

 any other class. 



